Prediction of load bearing capacity of piles from Statnamic pile testing data using neural networks

This paper explores the capabilities of neural networks to predict the static load bearing capacity of piles from rapid load pile testing data. A rapid load testing technique with the proprietary name of Statnamic testing is increasingly being used as an alternative to dynamic testing. It is less likely to cause damage to a pile than conventional dynamic methods, and removes the need for stress wave analysis and signal matching. Further-more, it is much faster and generally more economic than static tests. The main disad-vantages of rapid load tests are the problems associated with interpreting the test results in clay soils due to non-linear damping effects caused by variations in shear resistance with rate of shearing. A neural network is an information processing system, the archi-tecture of which essentially mimics the biological system of the brain. Artificial neural networks are ideally suited to assist engineers to interpret information from in-situ or laboratory measurements in complex problems. A back-propagation neural network has been trained using static and Statnamic load pile testing data. The trained network was then used to predict the static load-settlement behaviour from the rapid load data for other piles the data for which was not used in the training process.
The results of the prediction of the static load bearing capacity of piles using the neu-ral network have been compared with the field measurements. Comparison of the results show that artificial neural network can be used as an efficient engineering tool in predic-tion load bearing capacity of piles from rapid load testing data. These findings could en-able engineers to reduce the number of static load tests, thus saving time and cost.